Several studies have highlighted the burden on the families of mentally ill patients and so emphasised the need to investigate the effects of illness on family life and relationships. — A review of the literature showed that many of the published measures of the family lacked satisfactory reliability and validity. — A methodological study of 30 families in all of which there were children of school age or younger and in which one parent had newly attended a psychiatric facility is reported. For each family there was a 2–3 hour interview with the patient, a 3–4 hour interview with the spouse and a third interview of about an hour with the two of them together. At each interview there were two investigators who made independent ratings. There were different investigators for each of the three interviews so that there were 6 investigators and 6 sets of independent ratings for each family. — A basic distinction was made between concrete happenings or activities in the family, on the one hand and, on the other, feelings, emotions or attitudes concerning these events or the individuals participating in them. An interview designed to measure both types of variables is described. By the use of a particular kind of flexible and detailed questioning about recent events an attempt was made to free the measurement of events and activities from attitudinal biases. A different approach was used for the measurement of emotions. By the use of tape recordings and group discussions, interviewers were trained to recognise and rate differences in tone of voice as shown in the speed, pitch and intensity of speech. The third main type of scale consisted of overall summary ratings of relationships. — Each type of scale is illustrated by a number of examples for which detailed results are given. These are ‘warmth’ and ‘number of critical remarks’ as examples of ‘emotional’ scales, ‘participation of husband and wife in household tasks’ and ‘frequency of irritable acts’ as examples of scales of events or happenings, and ‘overall assessment of the marriage’ as a summary scale. Details are also given on the scale measuring ‘parent preferred by the child’ which does not readily fit into any of these 3 main categories. — High inter-rater reliability was demonstrated for the emotional scales. Respondent and rater biases were not found. By comparison of ratings made in different interviews, it was shown that similar feelings were expressed about the spouse in different situations, suggesting that the ratings were valid as well as reliable. A person's report of someone else's emotions, on the other hand, appeared to have a low level of validity in spite of good reliability. — Inter-rater reliability for the measures of activities and for the overall summary ratings was also high and, more important, it was shown that there was good agreement between the accounts of husbands and wives when interviewed separately. — A detailed examination was made of possible factors influencing husband-wife agreement. Slight biases associated with dissatisfaction and patient status were found but these were not sufficient to influence conclusions in most instances. Ways of reducing such bias are discussed.